skills.rest
microsoft
microsoftOFFICIAL
Explore all skills by this author →
Website𝕏@OpenAtMicrosoftGitHub

local-ai-agents

Implement local-first AI agents using Foundry Local and Qwen function-calling models.

73.6k|24.3k|Updated Nov 28, 2024
Search Tags:#mcp#Privacy#Foundry#local-ai#Qwen#slm#rAG
One-click install
npx skills add https://github.com/microsoft/ai-agents-for-beginners --skill local-ai-agents
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: local-ai-agents
Source: https://github.com/microsoft/ai-agents-for-beginners/tree/main/.agents/skills/local-ai-agents
Command: npx skills add https://github.com/microsoft/ai-agents-for-beginners --skill local-ai-agents

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires foundry_local, openai.

What problem does it solve?

This Skill addresses the need for privacy-preserving, offline-capable, and cost-effective AI agents that operate entirely on a developer's local workstation without relying on cloud inference.

Core Features & Use Cases

  • Local Orchestration: Run Small Language Models (SLMs) locally using Foundry Local to manage tool-calling loops.
  • Local RAG & MCP: Integrate Chroma for local vector storage and MCP servers for local tool execution over stdio.
  • Use Case: Build a private engineering assistant that searches local documentation and executes file-system tools without sending sensitive code or data to external cloud providers.

Quick Start

Use the local-ai-agents skill to initialize a Qwen2.5-7b-instruct model and connect an OpenAI-compatible client to the local Foundry endpoint.

Frequently Asked Questions about local-ai-agents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an AI agent fully offline?▼

Use Foundry Local to serve a model like Qwen2.5-7b-instruct on your machine. Configure your OpenAI client to point to the local endpoint provided by the Foundry manager to ensure all inference stays on-device.

What is the benefit of using Qwen for local agents?▼

Qwen models are specifically trained for function calling, which allows them to reliably generate structured tool calls. This capability is essential for transforming a standard chat model into an autonomous agent that can interact with local tools.

Can I use local RAG with this skill?▼

Yes, you can integrate Chroma as an in-process vector database. This allows your agent to upsert documents into a local collection and perform retrieval-augmented generation without external API calls.

What are the hardware requirements for local agents?▼

A minimum of 8 GB of RAM is recommended for running SLMs locally. While a GPU or NPU significantly improves performance, it is not strictly required for basic agentic tasks.

When should I avoid using local agents?▼

Avoid local agents for tasks requiring broad knowledge, long multi-hop reasoning, or massive scalability. For these scenarios, cloud-based models are more effective due to their larger parameter counts and distributed infrastructure.

Related Skills

Explore More Software Engineering Skills→
ClickHouseClickHouse
512

clickhousectl-local-dev

Provides tools and workflows for designing, validating, and preparing cloud-agnostic services for deployment.

Official
Advanced
zli5460zli5460
2

modal-serverless-gpu

Deploy GPU-accelerated machine learning models as serverless API endpoints.

Community
Intermediate
ModalityDanceModalityDance
1.1k

android-file

List, read, edit, and search files within Android development environments.

Official
Intermediate
iani-kuliiani-kuli

serving-llms-vllm

Deploy vLLM inference servers with quantization, tensor parallelism, and OpenAI-compatible APIs.

Community
Advanced